Cost of a human papillomavirus vaccination project, Zimbabwe

Abstract Objective To determine the cost of Zimbabwe’s human papillomavirus (HPV) vaccination demonstration project. Methods The government of Zimbabwe conducted the project from 2014–2015, delivering two doses of HPV vaccine to 10-year-old girls in two districts. School delivery was the primary vaccination strategy, with health facilities and outreach as secondary strategies. A retrospective cost analysis was conducted from the provider perspective. Financial costs (government expenditure) and economic costs (financial plus the value of existing or donated resources including vaccines) were calculated by activity, per dose and per fully immunized girl. Results The project delivered 11 599 vaccine doses, resulting in 5724 fully immunized girls (5540 at schools, 168 at health facilities and 16 at outreach points). The financial cost for service delivery per fully immunized girl was United States dollars (US$) 5.34 in schools, US$ 34.90 at health facilities and US$ 288.63 at outreach; the economic costs were US$ 17.39, US$ 41.25 and US$ 635.84, respectively. The mean financial cost per dose was US$ 19.76 and per fully immunized girl was US$ 40.03 (economic costs were US$ 45.00 and US$ 91.19, respectively). The largest number of doses delivered (5788) occurred during the second vaccination round (the second group’s first dose concurrently delivered with the first group’s second dose), resulting in the lowest financial and economic service delivery costs per dose: US$ 1.97 and US$ 6.79, respectively. Conclusion The mean service delivery cost was lower in schools (primary strategy) and when more girls were vaccinated in each round, demonstrating scale efficiency.


Introduction
Each year 266 000 women worldwide die of cervical cancer due to human papillomavirus (HPV) infection. 1,2 Cervical cancer is the fourth leading cause of estimated cancer deaths worldwide among women, most of which occur in low-and middle-income countries; the total is projected to increase to 416 000 deaths by 2035. 1,3,4 Low-and middle-income countries account for 84% (444 500 out of 527 600) of the world's cervical cancer burden. 2,4 HPV infection, one of the most common sexually transmitted diseases worldwide, is known as the main cause of cervical cancer, with HPV types 16 and 18 causing most cervical cancer cases. 5 The World Health Organization (WHO) recommends that all countries add the HPV vaccine to their national immunization programme, selecting a delivery strategy that is feasible with the current health infrastructure, affordable, cost-effective, sustainable and capable of achieving high coverage. 6 Zimbabwe is a low-income country with an estimated 4.9 million women aged 15 years and older at risk of developing cervical cancer. 5,7,8 Cervical cancer is the most frequent cancer among women (2270 new estimated cases of 8997 cancer cases in women each year) and the leading cause of morbidity from all cancers in Zimbabwe. 2 Zimbabwe's national health strategy and cancer prevention and control strategies mention HPV vaccination as a way to avert cervical cancer. 9,10 In 2013, Zimbabwe's Ministry of Health and Child Care proposed an HPV vaccination demonstration project supported financially by Gavi, the Vaccine Alliance. Gavi support for HPV vaccination demonstration projects was intended to allow countries to better understand the strategies and costs to deliver vaccine to a target population of girls aged 9-13 years. 11 Although Zimbabwe's project was initially designed with a three-dose schedule, it was revised to provide a twodose schedule following updated guidance issued by WHO in 2014. 12 To date, only a few published studies of the costs of HPV vaccination demonstration projects exist. [13][14][15][16] Our cost analysis adds original empirical retrospective data for delivering a two-dose vaccination schedule to two groups of girls in Zimbabwe, including the cost of vaccinating overlapping groups of girls (a second group's first dose concurrently delivered with a first group's second dose).
The existing literature on the cost of HPV vaccination demonstration projects has mostly focused on three-dose schedules and single cohorts. [13][14][15][16] The aim of this study was to determine the cost of Zimbabwe's two-dose HPV vaccination demonstration project. The study also contributes evidence on the costs of vaccinating overlapping groups of girls, providing insight into potential scale efficiencies. The study used a detailed empirical retrospective costing approach based on actual expenditure to understand the actual costs of delivering the vaccine.

Project implementation
The HPV vaccination project delivered vaccines to 10-yearold girls in two districts in 2014 and 2015 (Box 1). Zimbabwe selected school-based vaccination as the primary delivery strategy, with secondary delivery strategies at health facilities and outreach points (such as farms and missions) to reach girls

Selection criteria for districts
Peri-urban (two-thirds rural) districts Two major ethnic groups represented (Ndebele and Shona) High diphtheria-tetanus-pertussis vaccination coverage: > 90% Experience of schoolchildren vaccination programmes that would provide a potential platform for integrated delivery of HPV vaccine and other health interventions in the future

Target age group
10-year-old girls residing in the selected districts

Secondary delivery strategies
Health facilities; outreach points HPV: human papillomavirus. a The method of enumeration, based on school registers, likely underestimated the number of out-ofschool girls. Data sources: Government of Zimbabwe, 2013. 17 not attending school or who were absent during school vaccinations. Social mobilization was led by two local nongovernmental organizations (NGOs) and health ministry vaccinators to identify girls who missed vaccination. School health coordinators (teachers from the school trained in aspects of health) and school head teachers assisted in the logistics. These teachers scheduled vaccination dates, verified the eligibility of girls for vaccination by confirming birth dates, ascertained the fitness of those eligible for vaccination, reported any adverse events following immunization and facilitated referrals to the clinic if necessary. In each district, village health workers were engaged to offer support during the vaccination rounds, while a command centre facilitated logistics, monitored data collection and relayed information to the national level.
Both the selected districts in Zimbabwe implemented the project to reach an initial group of girls using three vaccination rounds ( Table 1). The target population for vaccination, based on school registers, was 2628 in Beitbridge district and 3880 in Marondera district. The first vaccination round attempted to reach the target population with the first dose. The second vaccination round delivered the second dose to the initial group of girls from the first vaccination round and the first dose to girls who missed the first vaccination round or who had recently turned 10 years old. A third vaccination round provided an opportunity to complete the vaccination series for girls from either of the previous rounds.

Study design
We used a retrospective, ingredientsbased approach to estimate the incremental, or additional, costs of imple-menting the HPV vaccination project. Costs were collected from the provider perspective, which included all costs of implementing the project from the government and donors, and were further categorized into financial and economic costs. Financial costs were defined as actual monetary payments or expenditure by the government, while economic costs were defined as financial costs plus the value of resources already paid for or owned by government or provided by other sources (opportunity cost), including the value of donated vaccines.
We based costs on the resources (ingredients) used for each major activity as defined by the WHO Cervical Cancer Prevention and Control Costing tool user guide. 18 Financial costs included expenditure by the health ministry. 18,19 Economic costs included financial costs plus expenditure by partners (e.g. vaccine and vaccine procurement by United Nations Children's Fund, funded by Gavi) and in-kind resource contributions from the health ministry and partners. We did not include evaluation costs for  18 For each activity, we calculated the value of the resources used based on actual expenditure, or the quantity of the resource used multiplied by that resource's unit cost and then summed across all resources. We estimated the daily value of personnel time using annual salary plus benefits by cadre divided by 222 working days per year; volunteer time was valued using a minimum wage estimate provided by the health ministry. 18,19 We valued inkind vehicle donations at a mean cost of the vehicle type based on new sale price (assuming no resale value, a useful life of 5 years and 222 working days per year), multiplied by the days the vehicle was used for the project. The vaccine purchase price per dose was United States dollars (US$) 4.60 (US$ 9.20 for a 2-dose vial); for analysis, we used the price of US$ 5.06 per dose procured after adding a 10% charge for airport clearance and transportation to the central medical stores. 11 Our cost analysis presents the empirical retrospective economic cost of vaccines actually used in the project, including doses delivered to the target population and doses wasted.
Since there were no major improvements made to the existing cold-chain infrastructure for the project (e.g. purchasing additional vaccine carriers or refrigerators), we calculated opportunity costs by using existing vaccine carriers and cold packs, as these were borrowed from routine vaccination sessions. We did not report the cost of vaccination waste disposal because the central medical stores and implementing districts were not able to provide an estimate of the share of waste disposal specific to HPV vaccination waste. Some aspects of service delivery costs (e.g. fuel) were reported by the health ministry as lumpsum costs. We divided these costs by the total number of vaccination sites reached in the vaccination round during which the expense was incurred to obtain cost per site, and then multiplied by the number of sites for each strategy (school, health facility or outreach) to obtain cost per strategy in the absence of information on resources used to reach each individual site.
We designated costs as introduction or recurrent costs based on the WHO Cervical Cancer Prevention and Control Costing tool user guide, 18 to delineate one-time investments from costs expected to continue on a recurring basis beyond the vaccine introduction phase. We defined introduction costs as investments needed during the initial phase of introducing a new vaccine into the immunization programme (i.e. investments in cold chain; social mobilization and information materials; microplanning; and training of supervisors, vaccinators and school staff). We defined recurrent costs as costs expected to continue on an ongoing basis (i.e. procurement of vaccines and related supplies; service delivery; and supervision, monitoring and evaluation). 18 This analysis considers the incremental resources needed to add HPV vaccination to an existing immunization programme. We excluded the costs of routine operation of the health system and immunization programme. 18 We collected costs in current prices from 2014, 2015 and 2016, adjusted for inflation to 2016 US$ using the Zimbabwe consumer price index. 20

Data collection and analysis
We collected the data through in-person consultations and by phone and email from 25 individuals from NGOs and the health ministry at the national, provincial, district and health-facility levels. Follow-up consultations were held with the same individuals to clarify and validate the initial information. We reviewed relevant programmatic and financial documents (e.g. annual reports, work plans, programme budgets and Zimbabwe's Gavi application 17 ) to triangulate with interview data and use as secondary data sources.
All resources recommended by the WHO Cervical Cancer Prevention and Control Costing tool were included in the data collection design for financial and economic costs by type of activity. Activities for which costs were reported in Zimbabwe are presented in Table 2. We conducted data analysis using Excel software, version 16.0 (Microsoft Corp., Redmond, United States of America) using the WHO Cervical Cancer Prevention and Control Costing tool user guide. 18 We estimated total financial and economic costs, then divided these by the number of doses and number of fully immunized girls to obtain the cost per dose and cost per fully immunized girl, respectively. We calculated introduction and recurrent costs using the cost of all doses delivered plus the cost of estimated doses wasted for groups in both districts. We calculated the service delivery category of costs separately by delivery strategy.

Results
The HPV vaccination project resulted in an estimated 11 599 doses delivered (11 251 at schools, 321 at health facilities and 27 at outreach points) and 5724 fully immunized girls (5540 at schools, 168 at health facilities and 16 at outreach points). The mean financial cost per dose for the overall project was US$ 19.76 and the mean economic cost per dose (which included the cost of the vaccine) was US$ 45.00 (Table 3). The mean financial and economic costs per fully immunized girl were US$ 40.03 and US$ 91.19, respectively.
The overall financial cost of the project was US$ 229 144 and the overall economic cost was US$ 521 946. Among activity categories, the share of financial cost was highest for social mobilization and information materials (24.1%; US$ 55 170) and lowest for vaccines, including vaccination-related supplies (0.1%; US$ 162; Table 4). The activity category with the highest share of economic cost was service delivery (21.7%; US$ 113 444) and the lowest was microplanning (9.6%; US$ 50 306).

Vaccination rounds
During the project, the mean number of girls vaccinated per school ranged between 12 and 36, which was higher than the mean number of girls vaccinated at health facilities (between 0 and 10) or outreach points (between 0 and 1; Table 5). The second vaccination round produced the largest number of doses delivered (5788; Table 6). By vaccination round, the lowest cost per dose was realized during the second round (US$ 1.97 financial cost; US$ 6.79 economic cost both in Marondera district; Table 6).

Service delivery costs
The service delivery costs varied by vaccination strategy. The mean service delivery costs per fully immunized girl for school vaccination, the primary strategy, were US$ 5.34 (financial cost) and US$ 17.39 (economic cost; Table 7). The mean financial and economic service The largest contributor to service delivery costs across all strategies was per diem payments (Table 7). These were paid to mobile teams of nurses (two to three in Beitbridge district; four in Marondera district) plus a driver, village health workers, school coordinators, command centre personnel (two in Beitbridge; three in Marondera), data collectors and one or two nurses at static health facility locations. For the school and outreach delivery strategies, per diem payments were paid to mobile vaccination teams. For the health facility delivery strategy, per diem payments were paid to health facility personnel who served as vaccinators.

Discussion
The service delivery financial costs per dose and per fully immunized girl were consistently higher in the health facility and outreach strategies. The higher costs are probably because both strategies were primarily geared towards supporting the school vaccination strategy and because of the small number of doses delivered in these settings (348 doses; 3% of the total). There were fewer outof-school girls than predicted and some of them were taken to be vaccinated at schools rather than via the expected health facility and outreach points. Costs for outreach and health facility strategies could be reduced if the size of mobile teams or command centre teams were reduced or if personnel were engaged without per diem payments. Any such changes in the composition of vaccination teams should be considered not only in terms of costs, but also the programmatic implications such as coverage and quality of the vaccination campaign. For school delivery, the financial cost per dose was lower when a larger number of girls was vaccinated in a single round (the second round, reaching overlapping groups of girls).
Introduction costs in this project may not be representative of national introduction costs in Zimbabwe since the country revised its plans for a three-dose vaccination series after preparations had begun, prompted by the change in WHO recommendations. 12 To change to a twodose schedule, information materials were re-printed, social mobilization sites were revisited and information had to be re-disseminated to communities, trainees and planning staff. This led to a higher than expected introduction cost for the project.
Zimbabwe's implementation of a two-dose schedule delivered to two overlapping groups of girls differs from the strategies used in all previously published HPV cost studies, [13][14][15][16] and the results are therefore not directly comparable. Nevertheless, comparison of findings from Zimbabwe with previous stud- In addition to differences in country context, implementation strategy and project structure, these cost differentials reflect differences in vaccination schedule (three-dose in previous projects versus two-dose in Zimbabwe). The findings from Zimbabwe reflect the cost to deliver HPV vaccine to two peri-urban districts and may not be directly comparable to settings represented by previous demonstration projects. For example, Zimbabwe's economic cost per fully immunized girl was US$ 91.19, including vaccine at US$ 4.60 per dose, while in the United Republic of Tanzania the economic cost per fully immunized girl in a rural setting was estimated at US$ 115.11 (presented in 2016 US$ 21 ), including vaccine at US$ 5 per dose. 15 A study of HPV vaccination projects found that as the amount of time to reach a school for vaccination increased, the cost subsequently increased. 16 Additionally, costs could differ in contexts with less robust routine immunization systems (as measured by diphtheriatetanus-pertussis coverage) than the two Zimbabwe project districts.
We found that the financial cost per dose decreased as the number of girls vaccinated increased. In the second round of vaccination when reaching overlapping groups of girls, the lowest financial costs per dose for vaccination service delivery were found when the most girls were vaccinated overall and on average per school. Other researchers found that the financial cost per dose decreased as the number of girls vaccinated increased; 16 however, their analysis did not include any countries with overlapping groups or cohorts. In our study, the third round of vaccination had the highest financial cost per dose, when the fewest girls were vaccinated and the fewest average number of girls were vaccinated per school. Others reported too that service delivery costs increased as fewer girls were vaccinated per school. 16 Our study has several limitations. The analysis was conducted retrospectively, requiring the HPV vaccination implementation staff to recall information. No written records were available for some information (e.g. personnel time); these estimates may therefore be subject to recall bias. Salary information  was given by cadre, not by specific personnel involved in programme activity; therefore, the salary levels assumed may be different from actual salaries.
This project was largely executed with Gavi support and supplemental donor funding, so our results do not permit conclusions about the costs of HPV vaccination in the absence of donor support. Furthermore, our findings are based on the cost analysis of a demonstration project and do not consider underlying programmatic factors such as the performance of the routine immunization programme. Other considerations not investigated as part of our cost analysis are also important for a country's decision on vaccination strategy. For example, the prevalence of human immunodeficiency virus (HIV) in the target population could affect decisions on HPV vaccine schedule, since a three-dose and not a two-dose schedule is recommended for people living with HIV. 6 Furthermore, the Zimbabwe project was designed primarily to target girls through schools, with health facilities and outreach points as secondary strategies. Future evaluations of HPV vaccination projects should consider a study design allowing for comparison of effectiveness and costs of different vaccination strategies, to determine the implications for optimal use of the health facility and outreach strategies, compared with the school vaccination strategy. Finally, it is important to note that Gavi demonstration projects only represent feasibility pilots of HPV vaccination in a few districts and do not offer estimates of nationally representative costs.
Following the demonstration project, Gavi approved financial support to the government of Zimbabwe to introduce the HPV vaccine nationally. 22 The first round of vaccination took place in May 2018 and aimed to reach 880 000 girls between the ages of 10 and 14 years (Manangazira P, personal communication, July 2018). Zimbabwe is the eighth country in Africa to introduce the HPV vaccine nationally. 22 In conclusion, this cost analysis provides new evidence regarding the resources required to deliver a two-dose vaccination schedule of the HPV vaccine using an overlapping delivery strategy to a new target population (adolescent girls) in Zimbabwe. As part of the country's first efforts to reach this population, the project provided an opportunity to understand the costs of various delivery strategies including health facility and outreach vaccination supplementing school-based vaccination to reach outof-school girls. The use of health facilities and outreach points as secondary vaccination strategies likely resulted in a higher mean service delivery cost for